Internal API update in preparation for Tinker integration #226
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Contributor
There was a problem hiding this comment.
Pull Request Overview
This PR updates internal APIs to support upcoming Tinker integration by enhancing token ID handling, logprobs tracing, and configuration flexibility.
Key changes:
- Added support for tracing and processing log probabilities from LLM responses
- Enhanced token ID extraction with null-safety checks and support for skipping empty token IDs
- Added
_add_return_token_idsconfiguration parameter to LLMProxy for conditional callback registration
Reviewed Changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| examples/.gitignore | Added Tinker-specific ignore patterns for logs and HTML reports |
| agentlightning/llm_proxy.py | Added _add_return_token_ids parameter to control token ID callback registration |
| agentlightning/instrumentation/agentops.py | Enhanced token ID extraction with null checks and added logprobs serialization |
| agentlightning/emitter/reward.py | Added TODO comment for tracer context improvement |
| agentlightning/algorithm/fast.py | Fixed exception handling order to check adapter existence before usage |
| agentlightning/adapter/triplet.py | Added span_to_triplet method and _skip_empty_token_spans parameter for filtering |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| filtered_llm_calls: List[Tuple[TraceTree, str]] = [] | ||
| for llm_call, agent_name in llm_calls: | ||
| triplet = self.span_to_triplet(llm_call.span, agent_name) | ||
| # This is a hot-fix for Tinker+CrewAI, which has some anonymous requests outside the trained agent. |
There was a problem hiding this comment.
Corrected spelling of 'hot-fix' to 'hotfix'.
Suggested change
| # This is a hot-fix for Tinker+CrewAI, which has some anonymous requests outside the trained agent. | |
| # This is a hotfix for Tinker+CrewAI, which has some anonymous requests outside the trained agent. |
| attributes["response_token_ids"] = list(first_choice.provider_specific_fields["token_ids"]) | ||
|
|
||
| # log probability | ||
| # This is temporarily. We need a unified conventions for classifying and naming logprobs. |
There was a problem hiding this comment.
Corrected 'temporarily' to 'temporary' and 'conventions' to 'convention' for grammatical accuracy.
Suggested change
| # This is temporarily. We need a unified conventions for classifying and naming logprobs. | |
| # This is temporary. We need a unified convention for classifying and naming logprobs. |
Contributor
Author
|
/ci |
|
✅ CI retrigger requested by @ultmaster. Fired |
Contributor
Author
|
/ci |
1 similar comment
Contributor
Author
|
/ci |
|
🚀 CI Watcher for correlation id-3452083391-mh9buyit triggered by comment 3452083391
✅ All runs completed. |
|
🚀 CI Watcher for correlation id-3452075952-mh9bva2a triggered by comment 3452075952
✅ All runs completed. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR includes several hot fixes in preparation for the Tinker example.
_add_return_token_idsin LLMProxy.Most API updates are very experimental (and underscored). They are only to make the Tinker integration work quickly.
We may need to rethink about: